• Computer Science > Artificial Intelligence [Submitted on 20 Feb 2026] Title:Cross-Embodiment Offline Reinforcement Learning for Heterogeneous Robot Datasets View PDF HTML (experimental)Abstract:Scalable robot policy pre-training has been hindered by the high cost of collecting high-quality demonstrations for each platform. • In this study, we address this issue by uniting offline reinforcement learning (offline RL) with cross-embodiment learning. • Offline RL leverages both expert and abundant suboptimal data, and cross-embodiment learning aggregates heterogeneous robot trajectories across diverse morphologies to acquire universal control priors. • We perform a systematic analysis of this offline RL and cross-embodiment paradigm, providing a principled understanding of its strengths and limitations. • To evaluate this offline RL and cross-embodiment paradigm, we construct a suite of locomotion datasets spanning 16 distinct robot platforms. • Our experiments confirm that this combined approach excels at pre-training with datasets rich in suboptimal trajectories, outperforming pure behavior cloning.
Article Summaries:
- A recent study proposes a new approach to pre‑training robot control policies by combining offline reinforcement learning (RL) with cross‑embodiment learning. Offline RL uses both expert and abundant suboptimal data, while cross‑embodiment aggregates trajectories from diverse robot morphologies to build universal control priors. The authors built a locomotion dataset covering 16 different robots and showed that the combined method outperforms pure behavior cloning when many suboptimal trajectories are present. However, as the proportion of suboptimal data and the number of robot types grow, conflicting gradients across morphologies hinder learning. To address this, they introduced a static grouping strategy that clusters robots by morphological similarity, applying group‑level gradients. This simple technique reduces inter‑robot conflicts and surpasses existing conflict‑resolution methods.
Sources: